An algorithm for computation of the scene geometry by the log-polar area matching around salient points

Authors:
Bogusław Cyganek
Affiliations:
AGH, University of Science and Technology, Kraków, Poland
Venue:
SOFSEM'08 Proceedings of the 34th conference on Current trends in theory and practice of computer science
Year:
2008

Citing 10
Cited 0

Multidimensional Orientation Estimation with Applications to Texture Analysis and Optical Flow

IEEE Transactions on Pattern Analysis and Machine Intelligence
3D motion recovery via affine epipolar geometry

International Journal of Computer Vision
In Defense of the Eight-Point Algorithm

IEEE Transactions on Pattern Analysis and Machine Intelligence
A Sequential Factorization Method for Recovering Shape and Motion From Image Streams

IEEE Transactions on Pattern Analysis and Machine Intelligence
Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography

Communications of the ACM
Multiple view geometry in computer vision

Multiple view geometry in computer vision
Digital Image Processing: PIKS Inside

Digital Image Processing: PIKS Inside
The Geometry of Multiple Images: The Laws That Govern The Formation of Images of A Scene and Some of Their Applications

The Geometry of Multiple Images: The Laws That Govern The Formation of Images of A Scene and Some of Their Applications
Invariant Fitting of Two View Geometry

IEEE Transactions on Pattern Analysis and Machine Intelligence
Image registration using log-polar mappings for recovery of large-scale similarity and projective transformations

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Computation of a scene geometry belongs to the one of the fundamental problems of computer vision. It can be computed from point correspondences found in a pair of stereo images or a video sequence. This is achieved by an image matching algorithm. However, found correspondences usually are burdened with large errors due to noise and outliers. In this paper we propose an improvement to the point matching algorithm which is twofold. At first the salient points are found which are corners detected by the structural tensor. Then the log-polar representations are computed around found salient points and the matching is done in the extended log-polar space. Such representation has very desirable properties which allow detection of a local change of scale and rotation of the matched areas. This feature is employed in the matching algorithm to eliminate outliers. The proposed method can be used in variety of computer vision tasks, such as stereovision, recovering shape and motion from video, or camera calibration and autocalibration.